Disclosure: I build AnkiQuiz, the card-based quiz tool discussed later in this guide. AnkiQuiz is not affiliated with Anki, AnKing, AnkiHub, UWorld, AMBOSS, or the NBME. It does not change Anki scheduling or replace validated question banks.
The best Anki settings for medical school are not a magic preset. They are settings that preserve the material you actually need while leaving time for lectures, clinical duties, sleep, and question-bank practice.
For most students using a current Anki release, a sensible starting point is FSRS enabled, desired retention left near the 90% default, short same-day learning steps, honest answer-button use, and a new-card limit based on measured review capacity. Then use your own statistics and the FSRS simulator to adjust workload.
The short answer
- Update every Anki client before enabling FSRS.
- Enable FSRS globally and keep learning and relearning steps shorter than one day.
- Start desired retention around 90%; simulate before moving toward 95–99%.
- Optimize parameters from your own history instead of copying someone else's.
- Press Again when you forgot; Hard is a passing grade.
- Set new cards per day from available time, not a classmate's card count.
- Avoid hiding an oversized queue behind a low maximum-review cap.
- Check monthly statistics and protected question-bank time before changing settings.
What FSRS changes
FSRS estimates each card's difficulty, stability, and retrievability from review history, then schedules toward a desired probability of recall. According to the official Anki deck-options guide , it can more accurately estimate forgetting than the legacy scheduler and help retain more material for the same study time.
FSRS does not decide what deserves a card, teach the underlying concept, or make every review honest. Poor cards, inflated grades, and excessive new-card intake still create poor results.
A conservative FSRS setup for medical students
Enable FSRS only after updating all clients
Confirm that desktop Anki, AnkiMobile or AnkiDroid, and any other client you use support FSRS. Sync before and after the change. Remove legacy custom-scheduling code if you previously installed an older FSRS setup.
Keep learning steps short
The current Anki manual recommends learning and relearning steps that can finish on the same day and remain shorter than one day. A small number of short steps is enough for most workflows; long multi-day steps prevent FSRS from controlling the interval as intended.
Same-day repetition is not a substitute for learning. If a new pathology or pharmacology card makes no sense, return to the source and understand it before repeating the same wording five times.
Begin near 90% desired retention
Desired retention is the probability of recalling a card when it becomes due. The 90% default is a reasonable starting point, not a universal medical-school rule.
Increasing it shortens intervals and increases reviews. The increase becomes steep as retention approaches 100%, so a jump from 90% to 95% can cost much more time than the five-point difference suggests. Use Anki's workload simulator first.
Optimize with your own history
Use Optimize after you have enough real reviews for the model to learn from. Do not paste an influencer's parameters into your preset: their material, grading, and memory history are not yours. The manual says monthly optimization is sufficient; constant optimization is unnecessary.
Separate presets can help when collections differ substantially in difficulty or use. Do not create a preset for every organ system unless it serves a real scheduling purpose.
Leave “reschedule cards on change” off initially
When first enabling FSRS, allowing future reviews to adopt the scheduler is usually less disruptive than immediately rewriting every due date. The manual warns that rescheduling can make many cards due at once. Back up and simulate before deliberately rescheduling a mature collection.
How to use the answer buttons
- Again: you did not recall the required answer or were wrong.
- Hard: you recalled correctly, but with substantial effort.
- Good: you recalled correctly with normal effort.
- Easy: the answer was clearly easier than expected.
The most damaging shortcut is pressing Hard after forgetting. FSRS interprets Hard as successful recall and may produce intervals that are too long. Grade what you produced before revealing the back, not how familiar the answer looks afterward.
New cards and maximum reviews
New cards create future reviews. Choose intake only after protecting fixed commitments and measuring how long due reviews take across a normal week. The cards-per-day guide gives a capacity-based method.
A low maximum-review limit can hide overdue work without making it disappear. If the queue exceeds available time, reduce new cards, suspend low-value material, repair slow leeches, and protect questions. During Step 1 dedicated or a shelf rotation, the best setting may be fewer new cards, not a more aggressive scheduler.
Use statistics before changing retention
Review monthly rather than reacting to one bad day. The Anki statistics guide defines true retention and explains that mature-card retention under FSRS should be near desired retention over time.
- Is mature true retention reasonably close to the target?
- How many minutes do due reviews consume?
- Are new cards creating a rising queue?
- Are repeated failures concentrated in a few leeches?
- Are reviews displacing course learning or question-bank analysis?
If recall is low because cards are ambiguous or you press the wrong buttons, raising desired retention treats the symptom. Repair behavior and card quality first.
Settings by stage of medical school
Preclinical courses
Use stable settings and adjust card intake to the curriculum. Learn a bounded topic, then unsuspend only matching AnKing or personal cards. Avoid changing retention every time an exam approaches.
Step 1 dedicated
Protect timed questions, answer review, and official assessments. Keep mature high-value cards and target documented gaps. An abrupt retention increase can create exactly the review surge you do not have time for. See the Step 1 dedicated workflow .
Clerkships and Step 2
Clinical schedules vary. Use a rotation-sized active set, keep daily review time bounded, and let question-bank misses identify reusable gaps. Do not preserve every Step 1 detail at the expense of shelf preparation and patient care.
Where AnkiQuiz fits
AnkiQuiz does not alter due dates, answer history, desired retention, or FSRS parameters. It provides a separate changed-prompt check from selected card text.
After one topic's cards are mature, select 15–25 permitted text cards and generate a short quiz. Exclude hints, tags, IDs, protected question-bank text, and sensitive patient information. Ask for mechanisms or distinctions supported by the cards, verify every generated answer, and use misses to inspect the source cards.
This can reveal a quality problem scheduler statistics cannot: a card may be “correct” because exact wording or layout gives the answer away. The card-to-quiz guide explains field selection and generation.
Questions medical students commonly ask
What desired retention should a medical student use?
Start near 90%, simulate the workload, and adjust from your own time and statistics. Higher is not automatically better, and values near 100% can create a disproportionate review burden.
How often should I optimize FSRS?
The current Anki manual says monthly is sufficient. Optimize from your own history and avoid manually editing or copying parameters.
Should maximum reviews per day be 9999?
The important point is that the limit should not silently conceal a growing due queue. A high limit keeps work visible, but it does not require sacrificing sleep or questions. If reviews are unsustainable, reduce intake and active scope.
Should I press Hard when I forgot?
No. Press Again. Hard tells FSRS that recall succeeded with difficulty.
Can AnkiQuiz improve my FSRS schedule?
No. AnkiQuiz does not reschedule cards. It can expose knowledge that only works under the original prompt, giving you evidence to rewrite or relearn a card.
Use settings to protect the whole study plan
A scheduler is working when it supports the rest of medical education, not when it maximizes daily review counts. Start conservatively, grade honestly, measure a full week, simulate before large changes, and solve poor cards or excessive scope before chasing a new preset.